Man of Honour
Yes all stock settings on the cards.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Skyrim black screen in SLI
- Thread starter Durzel
- Start date
Yes all stock settings on the cards.
Yeah it would be for sure. Personally I think Skyrim has a problem with Quad SLI or the drivers don't support it in Skyrim and it only works by chance. Disable the 2nd GPU so you're running normal SLI and see how it goes. If that works then disable V-SYNC and OC a little so the voltage gets driven up. If it doesn't crash then you know it's just a Skyrim + quad SLI problem.
If you want to test to see if the problem is related to dynamic voltage adjustment, you could always try setting your cards to stay at 1.175v whether idle or under load by using Precision X, clicking on 'Adjust Voltage', put it at the top, (1.175V) and then apply the setting. I'm not sure whether Afterburner also allows you to do this or not becuse the voltage slider is locked on the 6xx series.
If Skyrim is then stable under long playing sessions, it could be a problem with the dynamic voltage on one or both cards, although I would've thought you'd see this happen on other games and benchmarks too, especially ones that put that cards under heavy load, but at least it should answer the question for you.
Unless I'm mistaken the voltage adjustment in Precision X does nothing for 600-series cards. I think the BIOS/hardware controls the voltage with no user intervention possible.
Heaven didn't show instability because it was on 1.175v constant (because of no vsync and the amount of load Heaven puts on the cards, being a benchmark util). BF3 is similar, I play it without vsync anyway and it's already a taxing game on all Ultra settings.
Skyrim, perversely, using less juice to maintain 60FPS is what its having problems with - not full/heavy load. I could play Skyrim without vsync but the result of that is that I get far more FPS than I actually need, and the cards behave - noise & temperatures - as if I'm running a benchmark. I can't really live with that as a permanent fix, if indeed it works... though it's worth a try I guess.
For clarity - the problem is not at max clocks/max voltage, as Heaven ran for 9 hours in high ambient temperatures with no issues. The issue is somewhere between idle and max clocks
... or possibly a Skyrim specific problem with Quad SLI.
It's hard to explain I guess, but it's the only thing that is making any sense in my head at the moment.
Heaven didn't show instability because it was on 1.175v constant (because of no vsync and the amount of load Heaven puts on the cards, being a benchmark util). BF3 is similar, I play it without vsync anyway and it's already a taxing game on all Ultra settings.
Skyrim, perversely, using less juice to maintain 60FPS is what its having problems with - not full/heavy load. I could play Skyrim without vsync but the result of that is that I get far more FPS than I actually need, and the cards behave - noise & temperatures - as if I'm running a benchmark. I can't really live with that as a permanent fix, if indeed it works... though it's worth a try I guess.
For clarity - the problem is not at max clocks/max voltage, as Heaven ran for 9 hours in high ambient temperatures with no issues. The issue is somewhere between idle and max clocks
... or possibly a Skyrim specific problem with Quad SLI.It's hard to explain I guess, but it's the only thing that is making any sense in my head at the moment.
Last edited:
Man of Honour
Played Skyrim for 2 hours so far no problems. Will have another go tonight.
At the moment I am not using vsync. Where do you turn it on is it in the game or nvidia control panel.
At the moment I am not using vsync. Where do you turn it on is it in the game or nvidia control panel.
Unless I'm mistaken the voltage adjustment in Precision X does nothing for 600-series cards. I think the BIOS/hardware controls the voltage with no user intervention possible.
...
For clarity - the problem is not at max clocks/max voltage, as Heaven ran for 9 hours in high ambient temperatures with no issues. The issue is somewhere between idle and max clocks
You may well be right. I know that the 6xx series can't have it's voltages raised above 1.175 because of the lock, but I read this overclocking guide which is for the 670 cards a couple of weeks ago and under the 'Precision-X' section, it says that the 'Adjust Voltage' feature allows you to peg / force it to run at the maximum voltage supported by the card, (1.175V) while you're benchmarking / testing.
I just thought that this might allow you to put the voltage up to the same level that Heaven ran at even while the card isn't under full load in Skyrim without modifying your vsync settings. That way you'll know whether dynamic voltage has anything to do with the problems you're having.
Man of Honour
The setting I have got by default in nvidia control panel for vsync is
Use the 3D application setting
I can not find anywhere in the game to turn it on so I presume the above setting means it is off.
As for temps
The above temps are possibly higher than yours. The fans are set on their default auto setting.
Last edited:
Well I think I do have a faulty graphics card. Mass Effect 3 crashed on me 3 times in exactly the same way Skyrim does.
I think the reason BF3 doesn't crash is because a) I have vsync off and b) it pushes the cards far more. ME3, like Skyrim, barely uses the cards (the joys of console-focused textures).
One thing I forgot to mention was that when I get these crashes sometimes for a split second I see white squares on the screen. Tonight during ME3 they stayed on for longer, it is exactly the same as what this guy was experiencing on a GTX 570. He apparently fixed his with more voltage, which isn't an option for me.
I also noticed when ME3 was running GPU3 and GPU4 (card 2) seem to use more voltage for around the same load as GPU1 and GPU2 (card 1). I guess with the manufacturing process no two cores are the same so require differing amounts of voltage. Unfortunately in my case I think I have a card which is not getting enough voltage for the clocks its trying to use. I can really think of no other explanation.
So my question now is this really - is this worthy of an RMA direct to EVGA? And how do I know which card is definitely at fault? My gut feeling is it's the one that's GPU3 and GPU4, would that always be the card lower in the case?
I think the reason BF3 doesn't crash is because a) I have vsync off and b) it pushes the cards far more. ME3, like Skyrim, barely uses the cards (the joys of console-focused textures).
One thing I forgot to mention was that when I get these crashes sometimes for a split second I see white squares on the screen. Tonight during ME3 they stayed on for longer, it is exactly the same as what this guy was experiencing on a GTX 570. He apparently fixed his with more voltage, which isn't an option for me.
I also noticed when ME3 was running GPU3 and GPU4 (card 2) seem to use more voltage for around the same load as GPU1 and GPU2 (card 1). I guess with the manufacturing process no two cores are the same so require differing amounts of voltage. Unfortunately in my case I think I have a card which is not getting enough voltage for the clocks its trying to use. I can really think of no other explanation.
So my question now is this really - is this worthy of an RMA direct to EVGA? And how do I know which card is definitely at fault? My gut feeling is it's the one that's GPU3 and GPU4, would that always be the card lower in the case?
Associate
for the amount of cash you paid for the card I'd expect it to be hand collected by Kelly Brook let alone posted back.
But yeah, white squares definitely sounds like its borked. Sorry dude.
But yeah, white squares definitely sounds like its borked. Sorry dude.
If that was the service level I'd receive I would've sent it back already. At least twice.
Problem I have now is how do I determine which card is faulty?
If I run each card on its own the problem will probably not manifest because one card doing the work of two will run differently to running 2 cards, if that makes sense. In other words if the problem only shows up when the cards are only ~50% loaded per core, running either card on its own with the same games/tests will probably not produce crashes since the card will be running differently (doing more work). Does that make sense?
Interested to hear thoughts...
Soldato
Try running them individually and see what happens. I'm tempted to think based on what you've posted that there's an issue with the voltage going through to the card while under next to no load.
Are GPU 3 & 4 in use the same amount as 1 & 2 in Skyrim/ME3? On my iPhone and it's difficult to scroll back and keep what I've already typed so apologies if it's already posted.
I'd be surprised if the game can handle 3 and 4 GPU's to be fair. Try running BF3 with vysnc enabled and check your GPU usage and try to mimic Skyrim in that sense by lowering textures until it's barely utilised. Then see what happens. I'm sure that BF3 uses more than 2 GPU's whereas I'm not so sure about the compatability of those games with simpler engines.
The white squares does sound worrying but to save you the pain of back and forth RMA'ing you could try the above and see what happens. I think the key is under utilising the cards in different scenarios to isolate the issue.
Are GPU 3 & 4 in use the same amount as 1 & 2 in Skyrim/ME3? On my iPhone and it's difficult to scroll back and keep what I've already typed so apologies if it's already posted.
I'd be surprised if the game can handle 3 and 4 GPU's to be fair. Try running BF3 with vysnc enabled and check your GPU usage and try to mimic Skyrim in that sense by lowering textures until it's barely utilised. Then see what happens. I'm sure that BF3 uses more than 2 GPU's whereas I'm not so sure about the compatability of those games with simpler engines.
The white squares does sound worrying but to save you the pain of back and forth RMA'ing you could try the above and see what happens. I think the key is under utilising the cards in different scenarios to isolate the issue.
Associate
What you're suggesting is that the card that is faulty will only bork out when there is less load on it, because running in SLI shares the load. I see what you're getting at, but I would have thought that increased loads would cause crashes. It's like having to lie down as soon as you get back from work because you're day has been completely stress free. Never come across a problem like this before.
How do you test for that?
Well, if it was me, I would draw up a table, and test each card, in each slot, over a variety of benchmarks, on low and high settings.
Either that, or you're gonna have to buy 10 30" monitors and run both cards fully overclocked on Skyrim fully maxed out with every texture on the Nexus while BF3 benches in the background.
How do you test for that?
Well, if it was me, I would draw up a table, and test each card, in each slot, over a variety of benchmarks, on low and high settings.
Either that, or you're gonna have to buy 10 30" monitors and run both cards fully overclocked on Skyrim fully maxed out with every texture on the Nexus while BF3 benches in the background.
Last edited:
My gut feeling is this..
If the presumed-faulty card ran at a set voltage all of the time (e.g. 1.175v) then I don't think I would be having any problems, indeed running the cards at full load (Heaven benchmark) didn't produce any crashes. The problem is I have no control over voltages - they go up and down automatically according to load, from 0.988v to 1.175v.
What I think is happening is that one of the cards, can't tell which one really, has a very subtle fault whereby at low load it's setting a clock (which is also dynamic, the whole boost clock thing) but the voltage it's also setting for that clock isn't sufficient for it. For a split second it doesn't have enough voltage for the specific clock, and it locks up.
This is imo why the problem doesn't manifest with vsync off, or with one card (because one card has to do much more work running at the same resolution/settings), it only occurs when the cores have a medium (~50%) amount of load.
In Skyrim et al all 4 cores are used pretty much the same amount from what I can tell, but it seems GPU3 and GPU4 need more voltage to achieve the same speeds. GPU1 and GPU2 seem to stick on 1.050v whereas GPU3 shows 1.062v most of the time, and GPU4 1.070v (I think). I put that down to a difference in the manufacturing process really.
It is a very convoluted problem I'll admit but I honestly can't see how it can be anything else at this point. I have no other cards installed in my system, I have a PSU that is more than capable of providing the requisite amount of power, I even have an extra 6-pin PCI-E power cable plugged into the auxiliary power connector on the motherboard.
The fact Heaven ran for 9 hours at full load, whereas Skyrim and now Mass Effect 3 will crash at medium load & medium temperatures, strongly suggests to me that it's a voltage issue.
If the presumed-faulty card ran at a set voltage all of the time (e.g. 1.175v) then I don't think I would be having any problems, indeed running the cards at full load (Heaven benchmark) didn't produce any crashes. The problem is I have no control over voltages - they go up and down automatically according to load, from 0.988v to 1.175v.
What I think is happening is that one of the cards, can't tell which one really, has a very subtle fault whereby at low load it's setting a clock (which is also dynamic, the whole boost clock thing) but the voltage it's also setting for that clock isn't sufficient for it. For a split second it doesn't have enough voltage for the specific clock, and it locks up.
This is imo why the problem doesn't manifest with vsync off, or with one card (because one card has to do much more work running at the same resolution/settings), it only occurs when the cores have a medium (~50%) amount of load.
In Skyrim et al all 4 cores are used pretty much the same amount from what I can tell, but it seems GPU3 and GPU4 need more voltage to achieve the same speeds. GPU1 and GPU2 seem to stick on 1.050v whereas GPU3 shows 1.062v most of the time, and GPU4 1.070v (I think). I put that down to a difference in the manufacturing process really.
It is a very convoluted problem I'll admit but I honestly can't see how it can be anything else at this point. I have no other cards installed in my system, I have a PSU that is more than capable of providing the requisite amount of power, I even have an extra 6-pin PCI-E power cable plugged into the auxiliary power connector on the motherboard.
The fact Heaven ran for 9 hours at full load, whereas Skyrim and now Mass Effect 3 will crash at medium load & medium temperatures, strongly suggests to me that it's a voltage issue.
Associate
Can you force constant voltage in afterburner
There is a setting for voltage in EVGA Precision X (the equivalent to Afterburner, it behaves the same way) but it does nothing. Voltage is apparently locked out on 600-series cards, it is controlled by the BIOS/hardware - it goes up and down depending on load.
I am 100% convinced that if I could force the max boost voltage (1.175v) on both cards that I would have no crashes at all, after all Heaven ran for 9 hours like this.
I am 100% convinced that if I could force the max boost voltage (1.175v) on both cards that I would have no crashes at all, after all Heaven ran for 9 hours like this.
Soldato
Yeah - the same for my 680 - can't affect the voltage at all. Even changing it to 1175mV does nothing.
Associate
Well, in that case, the only option available to you, is fault find which card is doing it, and send it back.
Is there a util I can use that can apply a defineable level of load?
If I run Skyrim with one card with an unchanged configuration (i.e. same resolution, AA/AF, etc) then that card is going to be doing a lot more work than it was doing before, and likely won't crash.
Therein lies the rub, I don't think either card will fail when run in isolation unless I can somehow simulate the exact conditions Quad-SLI produces load/voltage wise.
If I run Skyrim with one card with an unchanged configuration (i.e. same resolution, AA/AF, etc) then that card is going to be doing a lot more work than it was doing before, and likely won't crash.
Therein lies the rub, I don't think either card will fail when run in isolation unless I can somehow simulate the exact conditions Quad-SLI produces load/voltage wise.
Soldato
GPU-Z